The present study aimed at estimating the diagnostic accuracy of plain CT scan in detecting surface morphological variations of livers in comparison with gross anatomical examination by using cadaveric livers. The prevalence of common variations by both the methods was also determined. In the current study, we used the same cadaveric livers that were removed and stored in 10% formalin for gross examination as well as CT scan. Common and clinically relevant surface morphological variations as detected during gross anatomical examination were compared with the CT findings. These were accessory fissures, accessory lobes, pons hepatis, and lobulated caudate and quadrate lobes. Sensitivity, specificity, PPV, NPV and LR is estimated for all common variations. Our findings revealed that CT scan has variable diagnostic precision in detecting various surface morphological variations in liver.
The sensitivity of CT scan in detecting AF is 82% but the specificity is only 41.67%. A higher sensitivity of CT in detecting AF will help the surgeon to plan segmentectomy based on these fissures and will also reduce surprises during surgery. Higher specificity will reduce unwanted operative intervention in cases of blunt trauma of abdomen. This should be kept in mind when a person with abdominal trauma with hemoperitoneum or a person with ascites posted for a hepatobiliary surgery undergoes a screening CT scan where the blood and peritoneal fluid may also alter and hinder with the visualisation of AF or traumatic fissure.[2] The radiological appearance of a traumatic fissure and an anatomical fissure (normal or accessory) of liver is not the same. Usually when there is a natural hepatic fissure – normal or accessory, the peritoneum invaginates into it and may contain variable amount of fat. The presence of fat is responsible for the radiolucent appearance of the fissure.[8] Thus, the fat that fills the fissure will help in easy detection. The lesser the amount of fat in the fissure, the less is the chance of visualization on CT. This explains why narrow and shallow fissures (less than 0.5cm) get unnoticed on CT scan. A normal fissure and true AF will also have smooth contours due to peritoneum dipping into the fissure. But in a traumatic or iatrogenic fissure there will be peritoneal discontinuity and no fat will be seen in the fissure and the margins of the fissure will not be smooth. The most common site of AF is superior surface of right lobe of liver. As the major reason for development of AF is diaphragmatic indentation during development, the presence of muscle fibres in the fissures will act as a guide for natural fissures in vivo.[6] Hence, anomalies of diaphragm like eventration and hernia may also affect the appearance of liver. Associated pathologies of pleura and base of lung will also alter the appearance of diaphragm and liver. Moreover, AF in many cases is associated with radicles of portal vein or hepatic artery in the depth of the fissures and can act as a guide during surgery.[4] Whereas a traumatic fissure will not have such an association. CECT can circumvent many of these problems.
In the current study, the blood vessels in the liver had collapsed post-mortem and hence could not act as a guide for AF. The intensity of liver tissue is lesser than in a living person due to lack of blood. This could explain the reason for low number of TP and hence a lower sensitivity. Placing the cadaveric liver in water resulted in widening of small fissures which trapped water and air bubbles that hindered the visualisation of the normal contour, fat content and peritoneal covering of the fissure. This could explain the reason for higher number of FP and hence a lower specificity. Another reason for larger number of FP and hence low specificity could be due to the presence of artificial fissures on the liver like injury during dissection or during storage of specimens which would have masqueraded as AF. Despite using cadaveric livers that had minimal trauma, the specificity was found to be low.
Discrepancy between cadaveric and CT observations has been noted in previous studies on liver.[2, 5] But these studies used cadaveric livers for gross examination and compared the prevalence and distribution of AF in CT scans of patients undergoing CT for other abdominal conditions. Schramek et al in his study in 2013 has used the same cadaveric specimen for gross examination as well as CT to estimate the prevalence of AF. They found that the presence of gas artifacts, perimortal changes and fixation artifacts made precise detection of details of organs difficult that resulted in a low prevalence of variations by CT.[8] This was noted in the current study as well.
The sensitivity and specificity of CT to detect AL like Riedel’s lobe, Beaver’s lobe and others was found to be 63.63% and 95.65% respectively. Bilobed, trilobed and multilobed caudate and quadrate lobes were seen in 27 livers during gross examination. But CT could detect only 5 and thus an extremely low sensitivity (18.52%) was detected. The major reason for low sensitivity in lobar anomalies could be attributed to slight changes in orientation of the organ when compared to in vivo in-situ liver, along with fixation and gas artifacts, and post processing effects. The images that are taken as 5mm slices, are later reconstructed into 1mm slices. In this process, small variations get smoothened and may appear as undulations of the surface or border which are not detected as variations and thus results in a low sensitivity. The radiologists should also keep these possibilities in mind, as such findings may pose surprises during hepatic surgeries. A pathology arising from an AL may thus get undiagnosed. A proper diagnosis of a supernumerary lobe is also necessary for follow up as a pedunculated AL may undergo torsion. [3]
With a sensitivity and specificity of more than 80% to detect PH, plain CT is a good investigation to detect this condition both in groove for IVC or FLT which were the two sites where PH was detected in our study. Post-test probability was also found to be high for detecting PH.